multiple conformations of gal3 protein drive the galactose ... · mechanism, rather employs a...

Allosteric Activation of Gal3 protein

1

Multiple Conformations of Gal3 Protein Drive the Galactose Induced Allosteric Activation

of the GAL Genetic Switch of Saccharomyces cerevisiae

Rajesh Kumar Kar♣1*

, Hungyo Kharerin¶1,2

, Ranjith Padinhateeri2 and Paike Jayadeva Bhat

#1

1Molecular Genetics Laboratory,

2Physical Biology Laboratory,

Department of Bioscience and Bioengineering,

IIT Bombay, Powai, Mumbai 400076, INDIA

♣ has performed all the experimental work

¶ has performed all the molecular dynamics simulations work

Current Address: Department of Biochemistry, College of Agricultural and Life Sciences, 433

Babcock Drive, University of Wisconsin-Madison, WI 53706

# Corresponding Author E-mail address: [email protected]

Running title: Allosteric Activation of Gal3 protein

Keywords: Error Prone PCR, Galactose signaling, Conformational heterogeneity, Allostery,

Molecular dynamics simulations.

Abbreviations: d.o.: drop out, 2-DG: 2-deoxy-galactose, Gal3p: Gal3 protein, EtBr: Ethidium

bromide, GAL3c:GAL3 constitutive mutant, ORF: open reading frame, EPPCR: Error

Prone PCR, CMD: Canonical molecular dynamics, TMD: Targeted molecular dynamics,

SO: Super-open

.CC-BY-NC-ND 4.0 International licenseIt is made available under a (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint. http://dx.doi.org/10.1101/032136doi: bioRxiv preprint first posted online Nov. 18, 2015;

mailto:[email protected]

http://dx.doi.org/10.1101/032136

http://creativecommons.org/licenses/by-nc-nd/4.0/


2

Abstract

Gal3p is an allosteric monomeric protein which activates the GAL genetic switch of

Saccharomyces cerevisiae in response to galactose. Expression of constitutive mutant of Gal3p

or over-expression of wild-type Gal3p activates the GAL switch in the absence of galactose.

These data suggest that Gal3p exists as an ensemble of active and inactive conformations.

Structural data has indicated that Gal3p exists in open (inactive) and closed (active)

conformations. However, mutant of Gal3p that predominantly exists in inactive conformation

and yet capable of responding to galactose has not been isolated. To understand the mechanism

of allosteric transition, we have isolated a triple mutant of Gal3p with V273I, T404A and N450D

substitutions which upon over-expression fails to activate the GAL switch on its own, but

activates the switch in response to galactose. Over-expression of Gal3p mutants with single or

double mutations in any of the three combinations failed to exhibit the behavior of the triple

mutant. Molecular dynamics analysis of the wild-type and the triple mutant along with two

previously reported constitutive mutants suggests that the wild-type Gal3p may also exist in

super-open conformation. Further, our results suggest that the dynamics of residue F237 situated

in the hydrophobic pocket located in the hinge region drives the transition between different

conformations. Based on our study and what is known in human glucokinase, we suggest that the

above mechanism could be a general theme in causing the allosteric transition.

Introduction

Information transfer through molecular recognition plays a vital role in cellular processes

such as catalysis, transcriptional regulation and differentiation. This occurs mainly through

allostery, wherein ligand binding at one site induces a conformational change at another site

(Monod et al. 1965; James & Tawfik 2003). This alteration in the conformation of the protein is

conveyed to the downstream target in a highly calibrated and quantitative fashion. Allostery was



http://dx.doi.org/10.1101/032136



3

originally thought to occur only in multimeric proteins either through the induced fit or through a

conformational selection mechanism as proposed in KNF (Koshland 1958) and MWC (Monod et

al. 1965) models respectively. Subsequent success of the structural studies in exploring the

MWC model suggested that allostery in general can be understood from a structural standpoint

(Marsh & Teichmann 2015). A model for explaining allostery without invoking a conformational

alteration was proposed based on thermodynamic considerations (Cooper & Dryden 1984).

While the above models provide a fundamental conceptual framework for our current

understanding, the allosteric behaviour of many proteins need not necessarily follow one or the

other mechanisms nor the postulates of the above models are adequate to explain the complex

allosteric behaviour of many proteins. For example, it has been suggested that strong and long-

range protein-ligand interactions favour induced fit model while weak and short range

interactions favour conformational selection mechanism (Okazaki & Takada 2008). On the other

hand, the rate of conformational transition could determine whether induced fit or the

conformational selection mechanism dominates (Hammes et al. 2009; Vogt & Di 2012).

Recently, computational analysis has indicated both conformational selection and induced fit

mechanisms operate during the allosteric transition of LAO proteins (Silva et al. 2011). In

contrast to the above studies, how a ligand acts as an agonist as well as an antagonist of the same

protein is difficult to explain using the KNF or MWC model (James et al. 2003; Motlagh &

Hilser 2012).

As opposed to the original postulate of MWC model wherein allosteric protein visits only

a few allowed conformations, it was suggested quite early on that allosteric proteins are likely to

visit a large number of conformations (Weber 1972). It has been suggested that allostery is

manifested due to the “redistribution of the conformational ensemble” in the presence of ligands

and therefore is an intrinsic feature of all dynamic proteins (Gunasekaran et al. 2004; Boehr et al.

2009; Blacklock & Verkhivker 2014). A large body of experimental evidence supports this



http://dx.doi.org/10.1101/032136



4

possibility of population shift (or conformational selection) induced by ligands. For example,

Nitrogen regulatory protein C (NtrC) is a signalling protein involved in the transcriptional

regulation of genes responsible for nitrogen assimilations in bacteria and its activation occurs

through a phosphorylation event (Kern et al. 1999). Using NMR, it has been demonstrated that

this protein exists as an equilibrium mixture of active and inactive states, indicating that

phosphorylation only shifts the equilibrium to the active state (Volkman et al. 2001). A similar

process of phosphorylation-mediated active-inactive conformational shift is also observed in

CheY, a monomeric single-domain response regulator protein involved in chemotaxis (Schuster

et al. 2001). Recent evidence suggest that CheY does not undergo a two-stage concerted

mechanism, rather employs a combination of the KNF and MWC mechanisms to function like a

switch (McDonald et al. 2012). In contrast to CheY, human glucokinase, a monomeric protein

with two distinct domains, has been shown to pre-exists in closed state prior to the introduction

of ligands, even though this state may be the minor species (Kamata et al. 2004; Zhang et al.

2006). Thus, although the induced fit mechanism is necessary, the conformational selection

appears to be the driving force for the allosteric transition.

Thus, at the conceptual level, thermodynamic, population shift, and structural basis of

allostery may appear disparate, but at the fundamental level, allostery needs to be explained from

all the above perspectives and accordingly a unified view of allosetry has been proposed (Tsai &

Nussinov 2014). Independent of the above, it has been proposed that few key residues are

associated with structural stabilization, allosteric couplings and correlated motions (Lockless &

Ranganathan 1999; Sinha & Nussinov 2001). Literature is replete with examples, wherein,

substitution of single key amino acid residues results in the stabilization of the active

conformation (or destabilization of the inactive conformation), which is otherwise attained only

after interacting with ligands (Kjelsberg et al. 1992; Blank et al. 1997). For example, in hamster

α-1b-adrenoceptor, substitution of Alanine at position 293 to any other residue results in the



http://dx.doi.org/10.1101/032136



5

constitutive activity (Kjelsberg et al. 1992). If mutations can stabilize the active conformation,

then, in principle, it should be possible to isolate mutations that stabilize the inactive state,

without altering the ability of the mutant protein to respond to its ligands, provided

conformational shift mechanism is an integral mechanism of allostery. To the best of our

knowledge, this corollary has not been experimentally verified. This is probably because, protein

in inactive conformations would normally be considered devoid of biochemical activity and

unlikely to be analysed further.

In S. cerevisiae, galactose activates the allosteric signal transducer Gal3p, which in turn

inactivates the repressor Gal80p, thereby activating the Gal4p dependent transcriptional

activation of the switch (Figure S1). The structure of Gal3p in apo as well as in holo (in

presence of galactose and ATP and Gal80p) forms has been solved at a resolution of 2.1 Å (Lavy

et al. 2012). In addition, an intermediate structure was observed during crystallization only when

the concentrations of the ligands used was low. Much before the structure was determined, it was

observed that, over-expression of Gal3p activated the GAL genetic switch even in the absence of

the inducer galactose (Figure S2C) (Bhat & Hopper 1992). Based on this, it was hypothesised

that Gal3p exists in active and inactive conformations and galactose merely increases the active

form of Gal3p (Figure S2). This hypothesis was subsequently supported by isolating mis-sense

mutations, which confer Gal3p, the ability to activate the GAL switch in the absence of galactose

(Figure S2B) (Blank et al. 1997). If the above hypothesis is true, then it should be possible to

isolate mutations that stabilize the protein in inactive conformation and yet such a mutant be able

to respond to galactose.

In this study, we report the isolation of a novel Gal3p mutant with V273I, T404A and

N450D substitutions, using a dual screening strategy. The ability of this Gal3p mutant to activate

GAL switch upon over-expression is very less as compared to the wild-type, but activates the



http://dx.doi.org/10.1101/032136



6

GAL switch in response to galactose. Genetic studies indicated that all the three substitutions are

essential for conferring the phenotype described above. Molecular dynamic studies, which

showed that unlike the wild-type Gal3p, the mutant form has higher propensity to exist in super-

open conformation. We discuss the implications of these observations in the context of

glucokinase, an allosteric protein, which has many functional overlaps with Gal3p and both seem

to follow a similar pathway of allosteric transition.

Materials and Methods

Yeast strains, media and growth conditions

Yeast strains (Table S1) were grown either in yeast extract, peptone and dextrose

(YEPD) or in synthetic medium supplemented with 3% (v/v) glycerol plus 2% (v/v) potassium

lactate (gly/lac), 2% glucose or 2% galactose (sigma, Germany) as carbon source. Whenever

required, Ethidium bromide (EtBr) and 3-amino-1, 2, 4-triazole (3-AT) (sigma, Germany) was

added to the medium to a final concentration of 20µg/ml and 10mM respectively. Yeast

transformation was done by lithium acetate mediated protocol (Gietz & Woods 2002). E. coli

transformation, plasmid amplification and isolation was carried out as with standard protocols

(Sambrook 1989).

Plasmid and Strain construction

Construction of PGAL1GFP (pYIPLac211GAL1GFP) and pRKK29 (pYIPLac204GAL1GFP)

A 1.5 kb cassette containing PGAL1GFP isolated from YCpGAL1-GFP (Stagoj et al. 2005)

as a PstI-EcoRI fragment was ligated at the corresponding sites of YIpLac211 to obtain

pGAL1GFP (pYIpLac211GAL1GFP) as described before (Kar et al. 2014). The same PstI-EcoRI

fragment was ligated at the corresponding sites of YIpLac204 to obtain pRKK29

(pYIpLac204GAL1GFP) (Table S2).



http://dx.doi.org/10.1101/032136



7

Strain construction

GAL3 disruption by KanMx4 cassette

GAL3 is disrupted in Sc723 (Blank et al. 1997) background by integrating KanMX4 cassette as

described before. ScRKK6 was generated by disrupting GAL3 with gal3KanMX4 cassette in

Sc723 (Kar et al. 2014).

GAL1 disruption by KanMx4 cassette

GAL1 was disrupted by integrating KanMX4. KanMX4 cassette was obtained by

amplifying genomic DNA of BY4742gal1:Kan by using primers PJB417

(5’GGATGGACGCAAAGAAGTTTAAT AATCATATTACATGGCATTACCACC3’) and

PJB418 (5’GTCCTTGTCTAACTTGAAAATTTTTC ATTGATGTCATACGAC3’). This strain

was obtained from EUROSCARF, in which GAL1 ORF is replaced by KanMX4 cassette. The

resulting PCR fragment contains KanMX4 flanked by 561bp from upstream and 537bp from

downstream of GAL1 ORF. Sc385 (gal3:LEU2) and Sc756 (gal1:URA3,gal3:LEU2) strains were

transformed with this PCR product and transformants were selected in YPD plate supplemented

with 200µg/ml of G418 and gives rise to ScRKK5 and ScRKK1 respectively. The putative

transformants were confirmed by phenotype analysis and PCR by primers PJB428 (5’CCA

GACCTTTTCGGTCACAC3’) and PJB427, which is designed 826bp upstream of GAL1 ORF

and 605bp internal of KanMX4.

Construction of ScRKK7 and ScRKK39

pGAL1GFP as well as pRKK29 was linearized with EcoRV located within the URA3 and

TRP2 locus respectively. This linearized pGAL1GFP plasmid was integrated at the ura3-52 locus

of Sc723 to obtain the ScRKK7 strain as described before (Kar et al. 2014). The linearized

pRKK29 was integrated at the trp2 locus of ScRKK1 to obtain ScRKK39 strain.



http://dx.doi.org/10.1101/032136



8

Error Prone PCR [EPPCR]

GAL3 ORF was cloned as a 1.6kb EcoRI and HindIII fragment into the corresponding site of

pYJM resulting in pYJM3 was available in the laboratory (Murthy & Jayadeva 2000). Error

Prone PCR [EPPCR] was done by using the classical protocol (Lin-Goerke et al. 1997). To

introduce random mutations in GAL3 ORF, EPPCR was carried out using the primers PJB328

(5’GGAAAGCGGGCAGT GAGCG3’) and PJB329 (5’GGCATGCATGTGCTCTGTATG3’) at

mutagenic condition (Table S3 and S4). Freshly prepared MnCl2 was always used for the

mutagenic condition.

Site directed mutagenesis

Site directed mutagenesis was performed by using “Quick change site-directed

mutagenesis protocol (QCM) developed originally by Stratagene. The oligonucleotides used for

site directed mutagenesis are listed in (Table S5). The mutations were confirmed by primer-

walking sequencing analysis from Scigenome as well as from Chromous biotech.

2-Deoxy-galactose (2-DG) toxicity assay

Constitutive expression of GAL1 in various transformants of gal3∆ cell was monitored

based on their ability to grow in gly/lac + 2-DG. It has been reported that cells constitutive for

galactokinase expression would not be able to grow in a media containing 2-DG due to toxicity

(Platt 1984). This toxicity is due to the accumulation of 2-DG-1p. Also, it has been reported that

2-DG does not substitute galactose for the signal transducing activity of Gal3p. Cells, which do

not produce galactokinase, do not show toxicity and therefore grow when provided with an

alternative carbon source glycerol plus lactate (gly/lac) (Kar et al. 2014).



http://dx.doi.org/10.1101/032136



9

Histidine growth assay

ScRKK1 (gal1∆gal3∆) cells were used for this assay, which has a PGAL1::HIS3 cassette

integrated at the LYS2 locus. ScRKK1 (gal1∆gal3∆) cells were pre-grown to mid-log phase in

complete medium containing gly/lac as the sole carbon source. Cells were appropriately diluted

and plated on to medium containing (a) uracil d.o. gly/lac (b) uracil d.o. gly/lac medium lacking

histidine with and without galactose. Medium lacking histidine contained 3-amino-1,2,4-triazole

(3-AT) to reduce background growth due to the leaky expression of HIS3 (Blank et al. 1997).

We have already used this assays to score the cells, which responds to galactose (Kar et al.

2014).

FACS analysis

Cells were grown in complete medium containing gly/lac and re-inoculated into medium

containing 2% glucose and 2% of galactose. Cells were harvested at different time intervals,

washed with PBS twice, re-suspended in PBS and kept in ice. A minimum of 50,000 cells were

analyzed for each data point. Flow cytometry data was collected using BD FACS ARIA-1 flow

cytometer with FITC filter for detection of GFP (Kar et al. 2014). Data were collected by BD

FACS DIVA software. The overlay and data analysis was done by FCS express 4 flow research

edition software (De Novo Software, Toronto, Ontario, Canada). At least three sets of

independent experiments were performed for FACS analysis. The FCS express 4 flow research

edition software overlaid data were tabulated (Table S6).

Electrophoresis procedure

Protein samples for SDS-PAGE were carried out according to the method of Laemmli (Laemmli

1970). 10% gels were used for all experiments. The electrophoresis was carried out at 15 mA per

gel.



http://dx.doi.org/10.1101/032136



10

Western blot analysis

Yeast cell extract was prepared by harvesting the cells grown in mid-log phase by centrifuging at

5,000 rpm for 5 min and then washed once with cold autoclaved double distilled water. Cells

were then resuspended in cold autoclaved Phosphate buffer saline. Protease inhibitor cocktail

and PMSF were added to the final concentration of 2.5 mM and 1 mM respectively. Equal

amounts of 0.45 mm diameter cold glass beads were added to each microfuge tube and subjected

to vortexing for 1 min, followed by incubation on ice for 1 min. This cycle was repeated for 10

times. Cell lysates were centrifuged at 13,000 rpm in a microcentrifuge for 25 min at 4°C. The

protein was quantified by BCA kit (Merck). The protein obtained was boiled with gel loading

buffer for 10 min and resolved in 10% SDS polyacrylamide gel. The protein was transferred

from the SDS gels to nitrocellulose membrane at 130 mA for 3 hour in transfer buffer. The

transferred protein was blocked with blocking buffer [PBS pH8.0 containing 3% (w/v) skimmed

milk powder and 0.02% (v/v) Tween20] for overnight. The blot was incubated for 1 hour in

1:1000 purified antibody for detection of Gal3 protein. The blot was then probed with the

secondary antibody conjugated with alkaline phosphatase at 1:5000 dilutions. Finally, blot was

developed in developing buffer containing NBT and BCIP. The protein bands were quantified by

Image J software (Schneider et al. 2012).

Protein models

The PDB coordinates of Gal3p, chain C of 3v2u (closed) and chain A of 3v5r (open) (Lavy et al.

2012) in which any ligand present was removed, were used as the starting structures for CMD

simulations in closed and open apo-states respectively. We retained ATP and galactose to

simulate the liganded structures. Since we do not have crystal structures for Gal3p mutants,

Gal3pIAD

, Gal3pD368V

and Gal3pF237Y

, we derived their structures by homology modeling with

crystal structures of Gal3p as templates. All modeling and fixing of missing residues were



http://dx.doi.org/10.1101/032136



11

carried out using the online version of homology modeler (Sali et al. 1995)

(http://toolkit.tuebingen.mpg.de/modeller).

Computational Details

All MD simulations were run with NAMD package (Phillips et al. 2005) using CHARMM force

fields (Brooks et al. 2009) and Tip3p (William L.Jorgensen 1983) water model. VMD

(Humphrey et al. 1996) “autoionize” package was used to neutralize the whole system. Solvent

box size was chosen in such a way that the minimum distance from the protein to the wall is

15Å. This will ensure there is no direct interaction with its own periodic images. Covalent bonds

involving hydrogen atom are made rigid by applying SHAKE. We have used CMD to study

equilibrium dynamics around native conformations. For all protein systems, energy was

minimized using conjugate gradient algorithm first for protein alone, then protein + water and

finally protein + water + ions. The solvated systems were heated to 300K and then equilibrated

as NPT ensemble in two steps. First, protein was fixed and the whole system was equilibrated for

1200ps. Second, the whole system was equilibrated for 1200ps with no constraint. The MD

integration time step used was 2.0 fs. In all the CMD simulations, coordinates were saved at

every 1ps interval for the analyses. Cα backbone root mean square deviation (RMSD) of

simulation trajectory as a function of time was plotted in the Figure S3A and B to assess

simulation stability and to check structural drift. To give enough relaxations to the modeled

residues (missing residues and mutation residues) in 3D environment, we set aside a good 20 ns

(shaded region in Figure S3A and B) as a relaxation step and the remaining last 100 ns trajectory

was selected to study equilibrium dynamics of proteins.

To study conformational transitions along activation pathway, we have employed TMD.

Here, an extra force given by the derivative of potential energy, UT M D = 0.5×k([RMSD(t) −

RMSD0 (t)]2)/N , is added to the existing force-field to drive the protein from open (start) to



http://toolkit.tuebingen.mpg.de/modeller

http://dx.doi.org/10.1101/032136



12

closed (target) conformation. N is the number of targeted atoms to which the force is applied and

k is the force constant in kcalmol−1

Å−2

unit. RMSD(t) is the instantaneous RMSD with respect to

the target structure, and RMSD0 (t) is a linearly decreasing targeted RMSD function where its

slope is fixed by the total simulation time (10 ns) and, the initial and final RMSDs between

starting and target structures. To check force constant and initial velocity dependencies, we

varied force constant k as F0=200 and F1=300 and initial velocities as V0, V1 and V2, and found

that force constant in that range does not show any significant change and also the simulation

does not depend on the initial velocity (Figure S3C). For our purpose, we have set k = 200

kcalmol−1

Å−2

. All simulations (CMD and TMD) (Table S7) were carried out at constant pressure

(1 bar) and temperature (NPT ensemble). The free energy, ∆G(t) = EMM + ∆Gsolv (t) − TS(t),

where EMM is the molecular mechanics energy from MD simulations, ∆Gsolv is the free energy of

solvation obtained using implicit solvent model, Poisson-Boltzmann Surface Area (PBSA)

method, T temperature in Kelvin, and entropy S calculated using Normal Mode Analysis

(Nmode), was calculated for every snapshot using an AMBER (Ambertools version 14.0)

module MM-PBSA-Nmode (Case et al. 2005). All the trajectory analyses were carried out using

trajectory analysis software grcarma (Koukos & Glykos 2013), AMBER (Ambertools version

14.0) and VMD. For visualization purposes Pymol and VMD were used.

Results

A wide range of over-expression of Gal3p turns on the GAL switch in the absence of

galactose

Before we set out to isolate mutations that stabilize Gal3p in inactive conformation, we

employed two complementary plate assays (Kar et al. 2014) to demonstrate the range of over-

expression of Gal3p that can cause constitutive expression of GAL genes. For this purpose, we

used two plasmid constructs, wherein GAL3 transcription is driven by CYC1 promoter (Figure



http://dx.doi.org/10.1101/032136



13

S4A) either from single-copy (pCJM3) or from multi-copy background (pYJM3). In one of the

assays, we spotted tenfold serially diluted cell culture of transformants of Sc385 (gal3∆ strain

bearing pCJM3 or pYJM3 on a Ura d.o. medium (to maintain the plasmid) containing glycerol

plus lactate (gly/lac) as the sole carbon source along with 2-deoxy-galatcose (2-DG) (See Table

S1 for genotype of strains and Table S2 for features of plasmids). If the over-expression of

Gal3p activates the GAL switch constitutively (i.e., in the absence of galactose), the

transformants cannot grow in the above medium. This is because, the constitutively expressed

galactokinase would convert 2-DG to 2-deoxy-galactose-1-phosphate (2-DG-1-P), which cannot

be further catabolised. Accumulation of 2-DG-1-P causes toxicity thereby preventing the

transformant from growing in gly/lac medium containing 2-DG. Transformants of gal3 bearing

pYJM3 do not grow on gly/lac plates containing 2-DG (Figure 1A) as demonstrated previously

(Das Adhikari et al. 2014). In this study, we demonstrate that gal3 strain transformed with

pCJM3, (single-copy Gal3p expression plasmid) is also unable to grow on gly/lac 2-DG plates

(Figure 1A) as compared to the vector control (pCJM). The second assay is dependent upon the

ability of the over-expressed Gal3p to activate the HIS3 expression from a PGAL1::HIS3 cassette

integrated in a ScRKK1 (gal1∆gal3∆his3∆PGAL1::HIS3) strain. Unlike the Sc385 (gal3∆) strain

used in the previous assay, ScRKK1 strain lacks the alternative signal transducer GAL1 and thus

allows us to test the constitutive activation of GAL genes upon over-expression of Gal3p in a

genetic context wherein only Gal3p over-expression activates the GAL switch. This strain

transformed with pCJM3 pYJM3 grow on gly/lac medium lacking histidine, while the strain

bearing the control plasmids pYJM or pCJM do not (Figure 1B). In addition to reproducing our

previous observation, we demonstrate that over-expression of Gal3p driven from CYC1 promoter

in a single-copy plasmid background also leads to the constitutive activation of GAL genes as

determined by monitoring the activation of GAL1 promoter (Figure 1).



http://dx.doi.org/10.1101/032136



14

Isolation of inactive mutant of Gal3p, which is responsive to galactose

It was reported that the stability of mutant forms of Gal3p is less as compared to the wild-type

(Platt et al. 2000). Based on this prior knowledge, we chose to introduce mutations in GAL3

ORF in the multi-copy plasmid background, that is, in pYJM3 (Figure S4A). The reason is that

in case the mutations introduced causes reduced stability, then over-expression of mutant

Gal3p’s from the multi-copy plasmid could compensate for the reduced stability and therefore be

able to score the phenotype of the mutant Gal3p. We introduced random mutations using error

prone PCR (EPPCR) in GAL3 ORF in pYJM3 (See Table S3 and S4 for EPPCR conditions) and

recovered the mutant plasmid populations as URA+ transformants of gal3∆ strain (Figure S4B).

These transformants were then replica plated on to Ura d.o. medium containing gly/lac plus 2-

DG to score for 2-DG toxicity as well as galactose plates containing EtBr to score for growth on

galactose. The transformants that grow on both the plates would contain the plasmids bearing the

GAL3 ORF with desired mutations.

The rational of the above protocol is as follows. Over-expression of those mutant Gal3p’s

that remain in inactive conformation would not constitutively activate the expression of

galactokinase and therefore would not cause toxicity. Such transformants would grow on gly/lac

plate containing 2-DG. Any mutation that knocks off the function completely would also be

picked up as a false positive in this screen. However, such mutants would not grow on the plates

containing galactose plus EtBr and hence can be easily weeded out (Figure S4B). Using this

dual screening strategy, we initially isolated ~ 50 transformants (Out of total 50,000

transformants) that showed the expected phenotype under both the assay conditions. Plasmids

were isolated from these yeast transformants, amplified in E.coli and retransformed into the

parent strain to confirm the phenotype. After going through this exercise several times to weed

out the false positive plasmids, we finally obtained three putative Gal3p mutant plasmid clones,



http://dx.doi.org/10.1101/032136



15

pYJM3mut1, pYJM3mut2 and pYJM3mut3. By employing a plasmid loss experiment strategy, we

confirmed that the phenotype is a plasmid borne trait and not due to genomic mutations (data not

shown). We also determined the identity of the plasmids bearing the Gal3p mutant ORF, using

restriction digestion analysis.

The above three putative plasmids were then subjected to 2-DG toxicity assay and His

growth assay. As expected, independent transformants of gal3∆ strain bearing pYJM3mut1,

pYJM3mut2 and pYJM3mut3 grew on gly/lac 2-DG plates, indicating that the above three

plasmids were unable to confer constitutive expression of GAL1 (Figure S5A). These

transformants grew on EtBr plates suggesting that they responded to galactose activation.

However, when these three plasmids were tested in His growth assay by transforming ScRKK1

(gal1∆gal3∆his3∆PGAL1::HIS3) strain, pYJM3mut1 and pYJM3mut2 were able to cause

constitutive expression of PGAL1::HIS3, which is not expected, while pYJM3mut3 failed to confer

constitutive expression as expected (Figure S5B). Thus, pYJM3mut3 does not confer constitutive

expression of GAL genes when tested under two different experimental conditions, while

pYJM3mut1 and pYJM3mut2 did not yield the expected phenotype. Since pYJMmut1 and 2 failed

to give the consistent phenotype in both the assays, we did not analyze these plasmids further.

The inability of pYJMmut3 plasmid to confer constitutive expression of GAL genes with

appropriate controls is shown (Figure 1).

Over-expression of mutant Gal3p does not result in the constitutive activation of GAL

genes

Since constitutive activation of GAL genes (i.e.in the absence of galatcose) is dependent on the

Gal3p concentration, it was important to establish the relative levels of wild-type Gal3p

expression driven by CYC1 promoter from a multi-copy (pYJM3) and single-copy plasmids

(pCJM3) as well as the mutant Gal3p expressed from multi-copy plasmids (pYJM3Mut3). To



http://dx.doi.org/10.1101/032136



16

determine this, western blot analysis was performed and the bands were empirically quantitated

(see Materials and Methods and Figure S6), based on the dilution of the protein samples and

using the loading control as the reference. Gal3p was found to be approximately 90 times more

in cell free extracts obtained from the transformant bearing multi-copy plasmid (pYJM3) as

compared to the single-copy plasmid (pCJM3) (Figure S6). Although the amount of Gal3p

expressed from single-copy plasmid (pCJM3) was less by approximately 90 times as compared

to the protein expressed from multi-copy (pYJM3), it was sufficient to give a clear cut read out of

constitutive expression using two independent plate assays (Figure 1). Western blot analysis

indicates that the steady state level of mutant Gal3p (pYJM3Mut3) was at least 10 times more

than what is produced from the single-copy background (pCJM3) (Figure S6). That is, wild-type

Gal3p expressed from single-copy (pCJM3), which is 10 times less than the mutant Gal3p

(pYJM3Mut3) activates GAL1 constitutively, as demonstrated by the plate assay (Figure 1).

Based on these results, we conclude that the mutant Gal3p is far less efficient than the wild-type

protein in constitutively activating the switch.

Three spatially well-separated substitutions are necessary to maintain Gal3p in inactive

and galactose responsive conformation

The GAL3 ORF present in pYJM3mut3 plasmid has V273I, T404A and N450D substitutions

(Figure S7, see Materials and Methods). To determine whether any of the single or double

mutations in any of the three combinations is sufficient to confer the unique property described

above, we generated the following plasmids with single (pYJM3V273I

, pYJM3T404A

and

pYJM3N450D

), or double mutations in all the three combinations (pYJM3IA

, pYJM3AD

and

pYJM3DI

) and a triple mutation (pYJM3IAD

) using SDM, for further analysis (See Table S5 for

primers for SDM in GAL3 ORF). The transformants bearing these mutant plasmids were

subjected to the plate assay as described previously. First, the transformant bearing pYJM3mut3



http://dx.doi.org/10.1101/032136



17

(original isolate, see Figure 1) and pYJM3IAD

(generated using SDM) exhibited the same

phenotype indicating that the phenotype is due to the mutations in GAL3 ORF and not due to

mutations elsewhere in the plasmids (Figure 1 and 2). Second, transformants bearing any of the

single mutations , pYJM3I, pYJM3

A and pYJM3

D as well as double mutations, pYJM3

IA, and

pYJM3DI

express GAL1 constitutively (Figure 2). This indicates that these mutants Gal3p’s

behave more like the wild-type Gal3p. However, transformant bearing pYJM3AD

was unable to

constitutively activate the switch but was not responsive to galactose unlike the pYJM3IAD

.

Quantification of the western blot analysis using densitometric scanning revealed that the

expression of Gal3pIAD

and Gal3pAD

were comparable (Figure 3A) while the Gal3pIAD

was at

least 5 times more than the WT Gal3p expressed from single-copy (pCJM3). The above data

indicate that Gal3pAD

is inactive and not responsive to galactose, while Gal3pIAD

is inactive but

responsive to galactose (Figure 2). Thus, Gal3pIAD

has a higher propensity than the wild-type

Gal3p to exist in inactive conformation. In addition to the phenotypic analysis (Figure 2), the

PGAL1::GFP expression was monitored in ScRKK39 (gal3∆gal1∆PGAL1::GFP) strain transformed

with the plasmids containing WT Gal3p from single-copy (pCJM3), Gal3pIAD

and Gal3pAD

expressed from multi-copy plasmid and the corresponding vector (pYJM) (Figure 3B). The

PGAL1::GFP expression was monitored in the transformants grown in gly/lac, in the absence of

galactose (reflects constitutive activation of the GAL switch) and in the presence of galactose

(reflects the responsiveness of Gal3p towards galactose). The mean fluorescence intensity of

gly/lac grown pYJM (vector control), pCJM3, pYJM3IAD

and pYJM3AD

transformants are 54.4,

76.96, 48.41 and 68.41, respectively. The mean fluorescence intensity of gly/lac + galactose

grown transformants are 56.84, 422.41, 376.04, and 63.55 (Table S6). As a control, we also

monitored the constitutive (gly/lac grown) PGAL1::GFP expression in the transformants bearing

pYJM3 (multi-copy plasmid carrying wild-type GAL3) and pYJM (multi-copy vector control). As



http://dx.doi.org/10.1101/032136



18

expected, the transformants carrying pYJM3 growing in gly/lac had higher GFP expression than

its corresponding vector control (Figure S8 and Table S6).

Thus, above results indicate that the constitutive PGAL1::GFP expression is lower in

pYJM3IAD

(Gal3pIAD

) as compared to pCJM3 (Figure 3B) even though the protein expression is

10 times more than the wild type Gal3p (Figure 3A) as confirmed from the western blot

analysis. Like wild type Gal3p, in response to galactose, Gal3pIAD

activates the switch, thus

suggesting that it predominantly exists in inactive conformation. Similar to Gal3pIAD

, Gal3pAD

is

also unable to activate the switch constitutively. However, unlike Gal3pIAD

, it is not responsive

to galactose (See Table S6 for statistical parameters of FACS analysis). Overall, these results

suggest that all the three substitutions (V273I, T404A and N450D) are essential to stabilize the

inactive conformation without sacrificing the ability of Gal3pIAD

to respond to galactose. We

carried out molecular dynamics analysis of wild type and mutant forms of Gal3p to further

elucidate the possible mechanism of the conformational transition.

Closed states of Gal3p show conformational bimodality

The crystal structure of Gal3p in closed conformation shows a lip-distance of ≈ 30 Å

between N-domain lip (D100) and C-domain lip (V375) (Figure S9). In the open conformation,

these lips open up to ≈ 40 Å. It has been suggested that the opening and closing movement

between these lips is related to the protein function (Menezes et al. 2003; Lavy et al. 2012). To

understand the role of amino acid substitutions in the conformational dynamics of the protein, we

have performed canonical molecular dynamics (CMD) simulations on both closed and open

conformations of WT Gal3p and mutants of Gal3p. In Figure 4, the distance between residues

D100 and V375 as a function of time (Figure 4A and C) and the corresponding conformational

distributions (Figure 4B and D) are shown.



http://dx.doi.org/10.1101/032136



19

In CMD starting with the closed structure (Figure 4A and B), the apo-forms of Gal3pWT

(WT Gal3p) and Gal3pIAD

(mutant of Gal3p with V273I, T404A, N450D substitutions) display

contrasting dynamics with respect to the inter-domain motion. The lip-distance for Gal3pWT

shows a bimodal distribution with one peak lying in the region ≈ 31-33 Å and the other peak in

the region ≈ 34-36 Å. In contrast, Gal3pIAD

is characterized by a broad distribution (spanning a

region of ≈ 35-41 Å) suggesting that the apo-form of Gal3pIAD

has the propensity to adopt open

conformations. However, in the presence of ligands the dynamics shift towards closed

conformation. The liganded form of WT (Gal3pWT

+Lig) shows a single peak. Interestingly,

Gal3pIAD

with ligand (Gal3pIAD

+Lig), on the other hand, behaves like the apo-form of Gal3pWT

,

showing bimodal distribution with similar nature of peak locations.

We also simulated the behaviour of the previously isolated constitutive mutants

(Gal3pD368V

and Gal3pF237Y

) which do not require any ligand for activation of GAL genes (Blank

et al. 1997). Both Gal3pWT

+Lig and Gal3pD368V

have the same range of lip-distance ≈ 31-33 Å.

In the case of Gal3pF237Y

, its highest peak falls in the region ≈ 34-36 Å (Figure 4A and B).

Therefore, we surmise that conformations with lip-distance ≈ 31-33 Å and 34-36 Å are

“Gal3pD368V

-like” (III) and “Gal3pF237Y

-like” (II) active conformations respectively, which bind

to Gal80p in the absence of ligands.

In CMD starting with an open conformation (Figure 4C and D), all the variants show

single peaks and interestingly all the mutants (Gal3pIAD

, Gal3pD368V

and Gal3pF237Y

) sample

structures with higher lip-distance than Gal3pWT

, especially Gal3pIAD

with lip-distance ≈ 44-46 Å

as opposed to Gal3pWT

≈ 40-42 Å. These structures are stable and well separated from the closed

structures, an indication that CMD can hardly cross high energy barriers. However, by starting

the simulation from different initial conformations, we can indeed obtain information about the

local dynamics and important insights in understanding the experiments. We classify structures



http://dx.doi.org/10.1101/032136



20

with lip-distance falling in the range ≈ 40-42 Å as “open” (I) and those falling in range ≈ 44-46

Å as “super-open” (SO). The justification for the conformational classification becomes apparent

from the results discussed in the following section.

Free energy of activation pathway reveals stable intermediate states

In order to access intermediate states and to complement our CMD results, we used targeted

molecular dynamics (TMD) simulations and sampled all possible structures starting from open-

state (lip-distance > 40 Å) to closed-state (lip-distance < 32 Å). To determine conformational

stability, free energy (see computational details in Materials and Methods) was calculated at

every time point. As the proteins undergo conformational transition along the activation

pathway, the lip-distances continue to decrease until they reach ≈ 7 ns, the time point from where

all the variants start to converge to ≈ 32-33 Å, except for Gal3pF237Y

, which converges to ≈ 36 Å

(Figure 5A). We call this region, which spans from 7 to 10 ns, the “region of stability” (shaded

region, Figure 5A). Remarkably, in the region of stability, all the structures seem to assume

single conformation with respect to lip orientation. We believe that using lip-distance as a large-

scale conformational order-parameter, our TMD has been successful in scanning all the possible

structures between open- and closed-states.

Free energy landscape along a potential transition pathway for the WT (Gal3pWT

) is

shown in Figure 5B. The result indicates that there are three important conformations: I (≈ 1 ns),

II (≈ 4 ns) and III (≈ 7-10 ns). Conformation I corresponds to the global minimum with lip-

distance ≈ 40-42 Å and it overlaps with the crystal structure of apo-form (open) of Gal3pWT

.

Conformation II represents a local minimum, which corresponds to a lip-distance of ≈ 34-36 Å

and possibly a stable intermediate state. We predict that this intermediate state is “Gal3pF237Y

-

like” active conformation. Structures lying in the olive shaded region (III) have lip-distance of ≈

31-33 Å and correspond to higher energy states in the absence of ligands (compare



http://dx.doi.org/10.1101/032136



21

conformations I, II and III in Figure 5B). We predict that this closed state is “Gal3pD368V

-like”

active conformation. Comparison of free energy landscapes for all the variants is shown in

Figure 5C. It can be seen that the initial structures in the first nanosecond (TMD ≈ 1 ns) are their

respective global minima (Figure 5C). As the TMD progresses along the activation pathway, the

free energy increases for all the variants. Strikingly, all the variants except Gal3pIAD

, show stable

intermediate states (minima), whereas for Gal3pIAD

, the intermediate states have higher energies.

In the region of stability (shaded region, Figure 5C), Gal3pIAD

structures have higher energies

compared to other variants, while Gal3pD368V

has a low stable minimum. Gal3pWT

and

Gal3pF237Y

, on the other hand, have intermediate energy levels. In Figure 5D, we present

superposition of structures taken from regions I, II, III (Figure 5B), and the global minima of

Gal3IAD

(Figure 5C), with respect to their C-domains. We found that between any consecutive

structures there is a rotation of ≈ 15° in their N-domains. Thus, the lip opening in super-open

state is 45° of rotation more than the closed state.

Residues T404A and N450D stabilize inactive conformation while V273I stabilizes active

conformation

It was previously demonstrated that F237, a hydrophobic hinge residue, plays a pivotal

role in stabilising the closed conformation of Gal3pF237Y

(Lavy et al. 2012). In the WT protein,

F237 is seated in the “hydrophobic pocket” formed by residues F247, M403, L65, F414, F418,

and I245 (Figure 6D; the hydrophobic pocket is indicated in grey color). It was proposed that

the spatial orientation of this residue directly governs the structural change of the protein, with a

solvent-exposed side-chain resulting in a closed conformation and its burial within the

hydrophobic pocket resulting in an open conformation (Lavy et al. 2012). In this study, solvent

accessible surface area (SASA) of residue 237 with respect to time was determined to study the

effects of mutations on its orientation (Figure 6A-C). It is clearly seen from the results that, in



http://dx.doi.org/10.1101/032136



22

closed conformer dynamics, residue F237 in Gal3IAD

tends to remain buried, whereas in the

Gal3pWT

, Gal3pD368V

and Gal3pF237Y

, it prefers to pop out into the solvent (Figure 6A).

However, in the open conformer dynamics, all the variants have residue 237 buried almost all the

time except for Gal3pF237Y

, which tends to remain exposed to solvent (Figure 6B). The

histogram of SASA of this residue also demonstrates that open (and super-open), intermediate

and closed conformations have SASA ≈ 15 Å2, 30 Å

2 and 50-60 Å

2 respectively (Figure S10).

SASA from TMD convincingly tells us that F237 in Gal3pIAD

prefers to remain buried (Figure

6C). We suggest that the T404A substitution, which lies close to the hydrophobic pocket,

extends the hydrophobic pocket and thus allows the F237 to get stably buried in Gal3pIAD

(Figure 6D). Additionally, the salt-bridge formation between residue Asp at position 450

(N450D) and residue Lys at position 488, constrains the relative motion between helices α15 and

α16 (Figure 6D and E). Thus, mutant of Gal3p with both the substitutions is unable to activate

the switch even in the presence of galactose (Figure 2A and B).

Root mean square deviations (RMSD) from TMD clearly show that both N- and C-

domains undergo significant local rearrangement with C-domain having higher degree of

rearrangement as compared to the N-domain along the activation pathway (Figure 7A and B). In

all the protein variants, the nature of RMSD evolution in N-domain is more or less the same.

Strikingly, the RMSD of C-domain for Gal3pIAD

is the highest and remain constant from ≈ 7 ns

onwards (Figure 7A shaded portion), suggesting the stabilizing role of V273I substitution when

the protein adopts active (closed) conformation. One of the major contributions to structural

rearrangement in C-domain may have come from the “polar-loop” fluctuation (Figure 7C and

D). This is evident from the RMSF calculation (Figure S11A and S12) showing that there is

higher fluctuation in open state than in closed state. In open state, this loop becomes more

solvent accessible and its conformational entropy increases. This conformational entropy is

minimized in closed state when the two lips come closer. This is important because residues 298



http://dx.doi.org/10.1101/032136



23

and 299 in the polar-loop are involved in interaction with Gal80p. Furthermore, the residue 299

forms a salt-bridge with residue 368 in all the protein variants, except for Gal3pD368V

,

irrespective of whether the global conformation is in closed or open state. Interestingly, in the

closed state, the loop adopts certain orientation in order to accommodate special structure of the

salt-bridge between 299 and 368 (Figure 7E; also see the snapshot at TMD = 8 ns). In

Gal3pD368V

, the residue 299 is perfectly trapped in the hydrophobic groove stabilized by D368V

substitution in the same manner that reduces conformational entropy of the loop (Figure 7F).

Discussion

While previous genetic and structural studies had suggested that wild-type Gal3p

can exist in active and inactive conformations (Bhat & Hopper 1992; Lavy et al. 2012), the

possible mechanism of allosteric transition was not deciphered. We decided to study the

allosteric transition by using molecular dynamics simulation (MD), as Gal3p is too large to study

using NMR. However, MD of only wild-type Gal3p and the constitutive mutants could be

performed and thus, our access was limited to only a part of the spectrum of conformations that a

wild-type Gal3p and its constitutive derivatives can potentially exist. Therefore, we were

constrained by the non-availability of mutant derivatives of Gal3p that would stabilize the

inactive conformation. To overcome this lacuna, we isolated Gal3pIAD

that appears to have high

propensity to exist predominantly in inactive conformation, which is at the other end of the

conformational spectrum. Using MD of mutant that we isolated (Gal3pIAD

), we were able to

capture a super-open (SO) conformation, in addition to I, II and III reported previously (Lavy et

al. 2012). The molecular dynamic analysis of the Gal3pIAD

is in consonance with conclusions

drawn based on the experimental results thus corroborating our molecular dynamic analysis. This

work has led us to unravel many unknown facets of allosteric transitions in Gal3p.



http://dx.doi.org/10.1101/032136



24

Although the positions of the three mutations are located far apart from each other and

away from the Gal80p interaction sites (Figure S11B), we can rationalize the importance of their

strategic location in stabilizing the inactive conformation without losing its ability to interact

with galactose. For example, the residue T404A lies adjacent to the hydrophobic pocket where it

harbors the side chain of residue F237. Substitution of residue T by hydrophobic residue A at

position 404 results in the formation of extended hydrophobic pocket, so that the side chain of

F237 gets stably buried in the hydrophobic pocket of Gal3pIAD

. Alteration of neutral charge

residue N to negatively charge residue D at position 450 leads to salt-bridge formation between

residue N450D and K488. An interesting consequence of such a perturbation in the local

interaction environment is that, it affects the relative motion between helices α15 and α16 via

conformational restriction (Figure 6D and E). Thus, it appears that the combined effect due to

co-substitution of T404A and N450D in the background of V273I substitution provides structural

stability to SO conformation in the absence of galactose. This could be attributed to the fact that,

Gal3pAD

exists predominantly in region I and in the region between I and II (~36-38Å) (Figure

S13B, C and D). Additionally, our MD and free energy calculations also reveal that Gal3pAD

also accesses region III (Figure S13A and D). Interestingly, Conformation III is immune to

galactose since the closed state does not allow the ligand to bind. Conformations with lip-

distance greater than 36Å may have low affinity for galactose. This may explain why Gal3pAD

does not response to galactose. Contrastingly, in Gal3pID

, the conformation with the minimum

free energy corresponds to region II (Figure S13A and D). Moreover, in Gal3pIA

and Gal3pID

,

V273I is a conservative substitution and thus may not contribute towards the stabilization of the

SO state. These results demonstrate that only these constellations of substitutions, V273I, T404A

and N450D, can give rise to the phenotype unique to Gal3pIAD

.

We suggest that the reason for the loss of the ability of Gal3pIAD

to activate constitutively

the GAL system is because, it exists mainly in SO conformation. Our analysis clearly indicates



http://dx.doi.org/10.1101/032136



25

that Gal3pWT

visits the closed (III), intermediate (II), and open (I) states more often than the SO

state. In other words, Gal3pWT

visits SO state very rarely. In contrast, in the Gal3pIAD

the SO

structure is the predominant form. It is interesting to note that three independent mutations are

essential to stabilize the SO structure that confers the unique property to the protein. This is in

sharp contrast to the stabilization of the closed structure that can be brought about by introducing

substitutions at single sites (Blank et al. 1997), which are spread across the primary sequence. It

is possible that mutations that can give rise to constitutive phenotype could be obtained by

destabilizing the inactive conformation as well. Unlike this, stabilizing the SO conformation

need a specific constellation of mutations.

Our analysis can be extended to explain the previously isolated constitutive mutations. A

mutation at the position 237 from F to Y (F237Y) confers the protein the ability to constitutively

activate the GAL switch (Blank et al. 1997). The position of this residue is located at the hinge

region and has been suggested that upon substitution of F by Y, the polar side chain of residue Y

interacts with the hydrophilic solvent thereby facilitating the protein to adopt active state (Lavy

et al. 2012). Consistent with this, we observe that the side chain of F237 remains buried in the

open (I) and SO structures, while it remains exposed to the solvent in the closed structures. Thus

the exposure of residue 237 to solvent is correlated with lip closure and vice versa. Our

simulation results suggest that mutation F237Y stabilizes the active intermediate states (II)

(Figure 5B).

Another constitutive mutation D368V located on the C-domain, which interacts with

Gal80p (Blank et al. 1997), is an example of mutations lying far away from the hinge region

(Figure S11B). Incidentally, the residue R299, which lies in the polar loop (LII) region (Figure

S9B) forms a salt-bridge with D368 (Figure 7E). We have shown that lip open-closure

movement and the structural local rearrangement due to salt-bridge interaction between residues



http://dx.doi.org/10.1101/032136



26

D368 and R299 in the C-domain, are correlated. However, the salt-bridge that stabilizes the

closed states (III) via reduction in fluctuations of polar loop (LII) region, is very different from

the salt bridge found in open (I) and super-open (SO) structures. Therefore, we surmise that, the

substitution of D by V, a hydrophobic residue, in D368V mutant, may stabilize the hydrophobic

surface and allow residue 368 to be perfectly localized in the hydrophobic groove. We provide

an explanation on the basis of structure and dynamics how mutation D368V stabilizes the closed

state (III) (Figure 5B).

In the light of our results, we propose a dynamic conformational model for allosteric

activation of Gal3p (Figure 8). The intrinsic fluctuation of F237 side chain has been shown to

correlate with domain-domain (lip) motion. When analyzing tertiary H-bonds networks

(excluding secondary structure H-bonds i.e. H-bonds involved in helices and sheets formation),

we found that the inter-domain H-bonds are not as extensive as compared to the intra-domain H-

bonds (Figure S14). This observation also supports the idea that the protein exhibits easy lip

movement. Thus, the protein can be modeled as a “hinge-wedge-pacman model” where the

residue F237 plays the role of a pivot-wedge at the hinge region. As shown in Figure 8, Gal3p

may preexist in four states defined by the order-parameters: lip-distance between the domains

and the SASA of F237. In the absence of galactose, wild-type Gal3p exists mostly in

energetically favorable open state (I). However, when galactose is present, it promotes

“conformational selection” on I, II and SO. We assume that galactose can bind to I, II and SO

except state III, but with decreasing affinity as the lip-distance increases. We surmise that

structure II has the highest affinity for galactose and once bound, this conformation may undergo

“induced-fit” to adopt more thermodynamically stable conformation III (closed state). The closed

structures, on the other hand, may be stabilized by ligand-mediated H-bonds and/or hydrophobic

interactions (residues: P112, V114, V201, V203, F259, A262, P263 and F373; (See movie S1)).



http://dx.doi.org/10.1101/032136



27

Therefore, we predict that the probable activation pathway may consist of I, II, III and SO traced

out on the parameter map defined by lip-distance and SASA of F237.

Human glucokinase, a monomeric bifunctional protein, is involved in the catalysis of

glucose phosphorylation as well as signal transduction. Similar to Gal3p, its interaction with a

downstream target protein is an integral part of signal transduction process (de, I et al. 2000).

Structural study has demonstrated that it exists predominantly in super-open (SO) state in the

absence of glucose (Kamata et al. 2004). Conformational restriction to open-closed states by a

small allosteric compound alters substrate saturation curve from an inherent “sigmoidal” to

acquired “hyperbolic” pattern (Kamata et al. 2004). Such a large-scale conformational transition

between SO and closed states with stable intermediate states has been shown to be responsible to

sigmoidal pattern of substrate saturation curve. Similar to what we observe in Gal3p, the

allosertic site in glucokinase also consists of a hydrophobic pocket (HP) (Huang et al. 2013)

made up of Tyr162, Val200, Ala201, Met202, Val203, Leu5451, Val42, Cal455 and Ala 456

with Ile159, which is analogous to F237 of Gal3p. As discussed before, a similar HP exists in

Gal3p, where the side chain of residue F237 is docked. Our results have demonstrated that

solvent accessibility of this side chain is correlated with lip motion. In closed state, HP is found

to be “inside-out” making it unfavorable for docking which causes residue 237 side chain to

remain exposed to the solvent. Whereas, in the open (super-open) state, the HP forms a deep

pocket favorable for docking.

Galactokinase of S. cerevisiae is a paraloge of Gal3p and can also transduce the galactose

induced signal of the GAL genetic switch. Galactokinase activity shows a hyperbolic saturation

pattern with respect to galactose (Timson & Reece 2002). These proteins share a high sequence

similarity, so much so, that corresponding residues V273, T404 and N450 of Gal3p are

conserved in Gal1p (Figure S7). If SO structure is responsible for sigmoidicity in glucokinase,



http://dx.doi.org/10.1101/032136



28

we predict that substitutions at equivalent positions by I, A and D should convert Gal1p from a

hyperbolic to a sigmoidal protein. It is intriguing to note that although glucokinase (hexokinase

family) and Gal3p (GHMP family) belong to different families, they seem to follow similar

mechanism of conformational transitions suggesting that such mechanisms are conserved and

could be found in other proteins as well.

To the best of our knowledge, there are two reports of the isolation of triple mutant that

confer a unique phenotype (Iiri et al. 1999; Alper et al. 2006). Of the two, the more relevant to

our study is the triple mutant of G mutant A366S, G226A and E268A(Iiri et al. 1999) , which

dominantly inhibits the receptor activation. None of the single or double mutation in any of the

combinations exhibits the negative dominant property. Interestingly, these mutations are highly

conserved and one of them is involved in the elimination of salt bridge. Salt bridges have been

shown to be responsible for conformational stability regardless of where they exist in the overall

protein structure (Kumar & Nussinov 1999; Bosshard et al. 2004). It appears that salt bridges in

general seem to have a vital role in allosteric transition. Thus, obtaining a unique conformation

with which a specific property is associated (in this case, Gal3pIAD

is inactive but responsive to

galactose) provides us a handle to study allostery resulting from protein dynamics. This attempt

highlights the importance of isolating mutations with unique properties and subjecting them to

detailed molecular dynamics simulation for elucidating the structural basis of allosteric

mechanisms.

Acknowledgements

We thank CRNTS of IITB for flow cytometry facility. We acknowledge the financial

support from IITB, CSIR, DST and 08BRNS004. We thank Dr. R. Komel for gifting

YCpGAL1GFP plasmid. We also thank YUVA PARAM supercomputing facilities, Pune, for

MD simulations purposes.



http://dx.doi.org/10.1101/032136



29

Conflict of interest

The authors declare that they have no conflicts of interest with the contents of this article.

Author Contributions

Wet lab experiments were conceived and designed by PJB and executed by RKK. MD

simulations were mainly conceived and performed by HK with input from RP. The paper was

mainly written by PJB and HK wrote all of the MD part. RP contributed in refining the

discussion. Performed all the experiments: RKK. Performed all the molecular dynamics

simulation: HK. Analyzed the data: RKK HK RP PJB. Contributed reagents/materials/analysis

tools: RP PJB.

Figures Legend

Figure 1. Over-expression of Gal3p driven by CYC1 promoter from multi-copy (pYJM3) or

single-copy plasmid (pCJM3) causes constitutive expression of GAL switch. (A) Overnight

cultures of the transformants of Sc385 (gal3 strain) bearing the indicated plasmids pre-grown

on gly/lac medium were serially diluted and spotted onto URA drop out gly/lac plates containing

2-DG. EtBr was used to allow the growth of only those cells that utilize galactose as the sole

carbon source and not the amino acids (Matsumoto et al. 1978). (B) Overnight cultures of the

transformants of ScRKK1 (gal3∆gal1∆his3∆PGAL1::His3) bearing the indicated plasmids, pre

grown on gly/lac medium were serially diluted and spotted onto the medium containing gly/lac

as the sole carbon source but lacking histidine and uracil (His and Ura d.o.). pYJM and pCJM are

the multi-copy and single-copy vector controls respectively. 3-amino-1,2,4-triazole (3-AT)

reduces the background growth due to the leaky expression of HIS3 (Blank et al. 1997).

Figure 2. Three mutations are required to confer the inactive conformation of Gal3p.

Transformants of (A) Sc385 (gal3∆) or (B) ScRKK1 (gal3∆gal1∆his3∆PGAL1::His3) strain



http://dx.doi.org/10.1101/032136



30

bearing pYJM3IAD

(triple mutant generated using site directed mutagenesis approach), pYJM3I,

pYJM3A, pYJM3

D(single mutants), pYJM3

IA, pYJM3

AD, pYJM3

DI (double mutants) pre-grown in

gly/lac were serially diluted and subjected to (A) 2-DG toxicity assay and (B). His growth assay.

Figure 3. Western blot analysis of Gal3ps and constitutive expression of the GAL switch.

(A) Cell free extracts obtained from gly/lac grown ScRKK5 (gal1∆gal3∆) transformant bearing

pYJM (vector control) pYJM3, pCJM3, pYJM3IAD

(triple mutant generated using site directed

mutagenesis approach), pYJM AD

(double mutants) were subjected to western blot analysis.

Zwf1p is used as the loading control. (B) Transformants of ScRKK39 (gal1∆gal3∆PGAL1::GFP)

bearing pYJM (vector control), pCJM3 (Gal3p over expressed from single-copy plasmid),

pYJM3IAD

(Gal3p triple mutant generated using site directed mutagenesis approach over

expressed from multi-copy plasmid) and pYJM3AD

(Gal3p double mutants over expressed from

multi-copy plasmid) were grown in gly/lac (shown in bright green, blue, red and orange color

respectively) and gly/lac+galatcose (shown in blue gray, turquoise, lime and black color

respectively) till mid-log. 50,000 cells were subjected to FACS analysis as described in material

and methods. The mean fluorescence intensity of GFP was measured in FITC channel of flow

cytometer. For the detail results of the FACS analysis refer Table S6. The experiment is repeated

at least thrice with different set voltage and all the three patterns were same, therefore one set of

data collected at a particular voltage is overlaid in this figure for comparison. (Figure 3B is a

part of figure S8 and is shown for clarity in comparing the overlaid mean GFP intensity of

indicated plasmid transformant).

Figure 4: Conformational equilibrium dynamics. (A, C) show time progression of distance

between N-domain lip (residue 100) and C-domain lip (residue 375) and (B, D) show their

corresponding distributions, respectively. (A, B) were simulated starting from closed



http://dx.doi.org/10.1101/032136



31

conformation while (C, D) were simulated using open conformation as the starting structure. The

grey dash lines in A and C indicate the corresponding distances found in X-ray crystal structures.

Figure 5: Allosteric activation pathway. (A) Time evolution of Lip-distance along activation

pathway. The grey dash lines indicate the corresponding distances in closed and open crystal

structures. (B and C) Conformational free energy along activation pathway. (B) The free energy

landscape for Gal3pWT

and (C) shows plot for all the variants where the global minima are

placed on the same energy level represented by black dash lines. The grey shaded regions

indicate minima along the pathway and labeled I and II. (A-C) The region of stability is shown

as shaded region in olive color and also labeled as III in (B). (D) Superposition of structural

representatives from regions I (yellow), II (green), III (blue) in (B) and the global minimum of

Gal3pIAD

(red) in (C) with respect to their C-domains: Conformation I, II and III are the open

“Gal3pWT

-like”, closed “Gal3pF237Y

-like” and closed “Gal3pD368V

-like” structures respectively,

and the conformation shown in red is the super-open “Gal3pIAD

-like” structure.

Figure 6: Role of T404A and N450D in stabilization of inactive conformations. SASA of

residue 237 with respect to time is presented in (A-C). (A) and (B) represent equilibrium

dynamics (CMD) and (C) represents biased dynamics (TMD). Since calculation of SASA of any

residue from interior protein is < 25 Å2, SASA ≤ 25 Å

2 is considered buried, whereas SASA >

25 Å2 is considered exposed to the solvent (black dash line). (D) Snapshots of CMD structures of

Gal3pWT

(green), in apo-closed (average SASA ≈ 50 Å2) and open (16 Å

2) states, and Gal3p

IAD

(purple) in open (15 Å2) state, showing different orientations of residue F237 (pink) seated in the

hydrophobic pocket formed by residues F247, M403, L65, F414, F418, and I245 (silver) in the

hinge region. In Gal3pIAD

, mutations T404A (red) forms extended hydrophobic pocket and



http://dx.doi.org/10.1101/032136



32

mutations N450D (red) forms salt-bridge with residue K488 (blue). (E) Distributions of angle

between helices α15 and α16.

Figure 7: Role of V273I in stabilization of active conformations. (A and B) Domain-RMSD

from TMD trajectory. (C and D) Polar-loop (LII) dynamics from CMD trajectory. Distance

between residues 299 (from LII) and 368 are plotted as a function of time. (E and F) TMD

snapshots at different time points. Hydrophobic residues are shown in surface rendition and the

polar residues are shown in cartoon representation. N-domain is colored blue and C-domain in

green. Residues 299 (orange) and 368 (pink) that forms salt-bridge in all the protein systems

(closed dynamics and open dynamics) except in constitutive mutant, Gal3pD368V

, are shown in

licorice representation. (E) Various structural forms of salt-bridge interaction between D368 and

R299. (F) Structural representative from a minimum lying in the olive-shaded region (Figure 5C

blue curve). In this mutant, residue D368V forms a part of the stable hydrophobic C-domain

surface.

Figure 8: Allosteric conformational model for Gal3p. Gal3p is represented as “hinge-wedge-

pacman” model where the hinge-wedge (triangle) depicts the residue F237 and the hydrophobic

lips (grey boxes) represent the residues: 112, 114, 201, 203, 259, 262, 263 and 373. Gal3p pre-

exists in four conformations: I, II and III (green) are energetically favourable structures and SO

(red) is energetically less favourable structure. The presence of galactose (yellow circle) induces

“conformational selection” on I, II and SO structures and encourages “induced-fit” between II

and III structures. We propose that galactose binds to II with high affinity and binds to SO with

low affinity (indicated by thickness of the arrows).



http://dx.doi.org/10.1101/032136



33

Literature Cited

Alper, H., J. Moxley, E. Nevoigt, G. R. Fink, and G. Stephanopoulos, 2006 Engineering yeast

transcription machinery for improved ethanol tolerance and production. Science 314:

1565-1568.

Bhat, P. J., and J. E. Hopper, 1992 Overproduction of the GAL1 or GAL3 protein causes

galactose-independent activation of the GAL4 protein: evidence for a new model of

induction for the yeast GAL/MEL regulon. Mol.Cell Biol. 12: 2701-2707.

Blacklock, K., and G. M. Verkhivker, 2014 Computational modeling of allosteric regulation in

the hsp90 chaperones: a statistical ensemble analysis of protein structure networks and

allosteric communications. PLoS.Comput.Biol. 10: e1003679.

Blank, T. E., M. P. Woods, C. M. Lebo, P. Xin, and J. E. Hopper, 1997 Novel Gal3 proteins

showing altered Gal80p binding cause constitutive transcription of Gal4p-activated genes

in Saccharomyces cerevisiae. Mol.Cell Biol. 17: 2566-2575.

Boehr, D. D., R. Nussinov, and P. E. Wright, 2009 The role of dynamic conformational

ensembles in biomolecular recognition. Nat.Chem.Biol. 5: 789-796.

Bosshard, H. R., D. N. Marti, and I. Jelesarov, 2004 Protein stabilization by salt bridges:

concepts, experimental approaches and clarification of some misunderstandings.

J.Mol.Recognit. 17: 1-16.

Brooks, B. R., C. L. Brooks, III, A. D. Mackerell, Jr., L. Nilsson, R. J. Petrella et al. 2009

CHARMM: the biomolecular simulation program. J.Comput.Chem. 30: 1545-1614.

Case, D. A., T. E. Cheatham, III, T. Darden, H. Gohlke, R. Luo et al. 2005 The Amber

biomolecular simulation programs. J.Comput.Chem. 26: 1668-1688.

Cooper, A., and D. T. Dryden, 1984 Allostery without conformational change. A plausible

model. Eur.Biophys.J. 11: 103-109.



http://dx.doi.org/10.1101/032136



34

Das Adhikari, A. K., M. T. Qureshi, R. K. Kar, and P. J. Bhat, 2014 Perturbation of the

interaction between Gal4p and Gal80p of the Saccharomyces cerevisiae GAL switch

results in altered responses to galactose and glucose. Mol.Microbiol. 94: 202-217.

de, l., I, M. Mukhtar, J. Seoane, J. J. Guinovart, and L. Agius, 2000 The role of the regulatory

protein of glucokinase in the glucose sensory mechanism of the hepatocyte. J.Biol.Chem.

275: 10597-10603.

Gietz, R. D., and R. A. Woods, 2002 Transformation of yeast by lithium acetate/single-stranded

carrier DNA/polyethylene glycol method. Methods Enzymol. 350: 87-96.

Gunasekaran, K., B. Ma, and R. Nussinov, 2004 Is allostery an intrinsic property of all dynamic

proteins? Proteins 57: 433-443.

Hammes, G. G., Y. C. Chang, and T. G. Oas, 2009 Conformational selection or induced fit: a

flux description of reaction mechanism. Proc.Natl.Acad.Sci.U.S.A 106: 13737-13741.

Huang, M., S. Lu, T. Shi, Y. Zhao, Y. Chen et al. 2013 Conformational transition pathway in

the activation process of allosteric glucokinase. PLoS.One. 8: e55857.

Humphrey, W., A. Dalke, and K. Schulten, 1996 VMD: visual molecular dynamics.

J.Mol.Graph. 14: 33-38.

Iiri, T., S. M. Bell, T. J. Baranski, T. Fujita, and H. R. Bourne, 1999 A Gsalpha mutant designed

to inhibit receptor signaling through Gs. Proc.Natl.Acad.Sci.U.S.A 96: 499-504.

James, L. C., P. Roversi, and D. S. Tawfik, 2003 Antibody multispecificity mediated by

conformational diversity. Science 299: 1362-1367.

James, L. C., and D. S. Tawfik, 2003 Conformational diversity and protein evolution--a 60-year-

old hypothesis revisited. Trends Biochem.Sci. 28: 361-368.

Kamata, K., M. Mitsuya, T. Nishimura, J. Eiki, and Y. Nagata, 2004 Structural basis for

allosteric regulation of the monomeric allosteric enzyme human glucokinase. Structure.

12: 429-438.



http://dx.doi.org/10.1101/032136



35

Kar, R. K., M. T. Qureshi, A. K. Dasadhikari, T. Zahir, K. V. Venkatesh et al. 2014 Stochastic

galactokinase expression underlies GAL gene induction in a GAL3 mutant of

Saccharomyces cerevisiae. FEBS J. 281: 1798-1817.

Kern, D., B. F. Volkman, P. Luginbuhl, M. J. Nohaile, S. Kustu et al. 1999 Structure of a

transiently phosphorylated switch in bacterial signal transduction. Nature 402: 894-898.

Kjelsberg, M. A., S. Cotecchia, J. Ostrowski, M. G. Caron, and R. J. Lefkowitz, 1992

Constitutive activation of the alpha 1B-adrenergic receptor by all amino acid

substitutions at a single site. Evidence for a region which constrains receptor activation.

J.Biol.Chem. 267: 1430-1433.

Koshland, D. E., 1958 Application of a Theory of Enzyme Specificity to Protein Synthesis.

Proc.Natl.Acad.Sci.U.S.A 44: 98-104.

Koukos, P. I., and N. M. Glykos, 2013 Grcarma: A fully automated task-oriented interface for

the analysis of molecular dynamics trajectories. J.Comput.Chem. 34: 2310-2312.

Kumar, S., and R. Nussinov, 1999 Salt bridge stability in monomeric proteins. J.Mol.Biol. 293:

1241-1255.

Laemmli, U. K., 1970 Cleavage of structural proteins during the assembly of the head of

bacteriophage T4. Nature 227: 680-685.

Lavy, T., P. R. Kumar, H. He, and L. Joshua-Tor, 2012 The Gal3p transducer of the GAL

regulon interacts with the Gal80p repressor in its ligand-induced closed conformation.

Genes Dev. 26: 294-303.

Lin-Goerke, J. L., D. J. Robbins, and J. D. Burczak, 1997 PCR-based random mutagenesis using

manganese and reduced dNTP concentration. Biotechniques 23: 409-412.

Lockless, S. W., and R. Ranganathan, 1999 Evolutionarily conserved pathways of energetic

connectivity in protein families. Science 286: 295-299.



http://dx.doi.org/10.1101/032136



36

Marsh, J. A., and S. A. Teichmann, 2015 Structure, dynamics, assembly, and evolution of protein

complexes. Annu.Rev.Biochem. 84: 551-575.

Matsumoto, K., Toh-e A, and Y. Oshima, 1978 Genetic control of galactokinase synthesis in

Saccharomyces cerevisiae: evidence for constitutive expression of the positive regulatory

gene gal4. J.Bacteriol. 134: 446-457.

McDonald, L. R., J. A. Boyer, and A. L. Lee, 2012 Segmental motions, not a two-state concerted

switch, underlie allostery in CheY. Structure. 20: 1363-1373.

Menezes, R. A., C. Amuel, R. Engels, U. Gengenbacher, J. Labahn et al. 2003 Sites for

interaction between Gal80p and Gal1p in Kluyveromyces lactis: structural model of

galactokinase based on homology to the GHMP protein family. J.Mol.Biol. 333: 479-492.

Monod, J., J. WYMAN, and J. P. CHANGEUX, 1965 ON THE NATURE OF ALLOSTERIC

TRANSITIONS: A PLAUSIBLE MODEL. J.Mol.Biol. 12: 88-118.

Motlagh, H. N., and V. J. Hilser, 2012 Agonism/antagonism switching in allosteric ensembles.


Murthy, T. V., and B. P. Jayadeva, 2000 Disruption of galactokinase signature sequence in gal3p

of Saccharomyces cerevisiae does not lead to loss of signal transduction function.

Biochem.Biophys.Res.Commun. 273: 824-828.

Okazaki, K., and S. Takada, 2008 Dynamic energy landscape view of coupled binding and

protein conformational change: induced-fit versus population-shift mechanisms.


Phillips, J. C., R. Braun, W. Wang, J. Gumbart, E. Tajkhorshid et al. 2005 Scalable molecular

dynamics with NAMD. J.Comput.Chem. 26: 1781-1802.

Platt, A., H. C. Ross, S. Hankin, and R. J. Reece, 2000 The insertion of two amino acids into a

transcriptional inducer converts it into a galactokinase. Proc.Natl.Acad.Sci.U.S.A 97:

3154-3159.



http://dx.doi.org/10.1101/032136



37

Platt, T., 1984 Toxicity of 2-deoxygalactose to Saccharomyces cerevisiae cells constitutively

synthesizing galactose-metabolizing enzymes. Mol.Cell Biol. 4: 994-996.

Sali, A., L. Potterton, F. Yuan, V. H. van, and M. Karplus, 1995 Evaluation of comparative

protein modeling by MODELLER. Proteins 23: 318-326.

Sambrook J.F.E.F.a.M.T. Molecular cloning: a laboratory manual . (2nd

edition). 1989.

Newyork, Cold Spring Harbor laboratory press, Cold Spring Harbor .

Ref Type: Serial (Book,Monograph)

Schneider, C. A., W. S. Rasband, and K. W. Eliceiri, 2012 NIH Image to ImageJ: 25 years of

image analysis. Nat.Methods 9: 671-675.

Schuster, M., R. E. Silversmith, and R. B. Bourret, 2001 Conformational coupling in the

chemotaxis response regulator CheY. Proc.Natl.Acad.Sci.U.S.A 98: 6003-6008.

Silva, D. A., G. R. Bowman, A. Sosa-Peinado, and X. Huang, 2011 A role for both

conformational selection and induced fit in ligand binding by the LAO protein.

PLoS.Comput.Biol. 7: e1002054.

Sinha, N., and R. Nussinov, 2001 Point mutations and sequence variability in proteins:

redistributions of preexisting populations. Proc.Natl.Acad.Sci.U.S.A 98: 3139-3144.

Stagoj, M. N., A. Comino, and R. Komel, 2005 Fluorescence based assay of GAL system in

yeast Saccharomyces cerevisiae. FEMS Microbiol.Lett. 244: 105-110.

Timson, D. J., and R. J. Reece, 2002 Kinetic analysis of yeast galactokinase: implications for

transcriptional activation of the GAL genes. Biochimie 84: 265-272.

Tsai, C. J., and R. Nussinov, 2014 A unified view of "how allostery works". PLoS.Comput.Biol.

10: e1003394.

Vogt, A. D., and C. E. Di, 2012 Conformational selection or induced fit? A critical appraisal of

the kinetic mechanism. Biochemistry 51: 5894-5902.



http://dx.doi.org/10.1101/032136



38

Volkman, B. F., D. Lipson, D. E. Wemmer, and D. Kern, 2001 Two-state allosteric behavior in a

single-domain signaling protein. Science 291: 2429-2433.

Weber, G., 1972 Ligand binding and internal equilibria in proteins. Biochemistry 11: 864-878.

William L.Jorgensen J.C.J.D.M.R.W.I.a.M.L.K. Comparison of simple potential functions for

simulating liquid water. J.Chem.Phys. 79[3], 926-935. 1983.

Ref Type: Journal (Full)

Zhang, J., C. Li, K. Chen, W. Zhu, X. Shen et al. 2006 Conformational transition pathway in the

allosteric process of human glucokinase. Proc.Natl.Acad.Sci.U.S.A 103: 13368-13373.



http://dx.doi.org/10.1101/032136



39

Figure 1 :

Ura d.o.Gal+EtBrUra d.o.gly/lac

+2-DG

Ura d.o. gly/lac

ga

l3∆

pYJM

pYJM3

pCJM3

pCJM

pYJM3 mut3

100 10-1 10-2 10-3 100 10-1 10-2 10-3 100 10-1 10-2 10-3

100 10-1 10-2 10-3 100 10-1 10-2 10-3 100 10-1 10-2 10-3

Ura d.o. gly/lac Ura His d.o.

gly/lac+gal+3-AT

Ura His d.o.

gly/lac+3-AT

ga

l3∆

ga

l1∆

his

3∆

A

BpYJM

pYJM3

pCJM3

pCJM

pYJM3 mut3



http://dx.doi.org/10.1101/032136



40

Figure 2 :

Ura d.o.Gal+EtBrUra d.o.gly/lac

+2-DG

Ura d.o. gly/lac

Ura d.o. gly/lac Ura His d.o.

gly/lac+gal+3-AT

Ura His d.o.

gly/lac+3-AT

100 10-1 10-2 10-3 100 10-1 10-2 10-3 100 10-1 10-2 10-3

100 10-1 10-2 10-3 100 10-1 10-2 10-3 100 10-1 10-2 10-3

pYJM3 IAD

pYJM3 I

pYJM3 A

pYJM3 D

pYJM3 IA

pYJM3 AD

pYJM3 DI

pYJM3 IAD

pYJM3 I

pYJM3 A

pYJM3 D

pYJM3 IA

pYJM3 AD

pYJM3 DI

ga

l3∆

ga

l3∆

ga

l1∆

his

3∆

A

B



http://dx.doi.org/10.1101/032136



41

Figure 3 :

FITC-A

Count

100

101

102

103

104

105

0

141

282

422

563B

Vector control

pCJM3

pYJM3 IAD

pCJM3

pYJM3 IAD

pYJM3 AD

pYJM3 AD gly

/la

c+ g

al

gly

/la

c

pY

JM

3 IA

D

pY

JM

3 A

D

pY

JM

3

pC

JM

3

pY

JM

3 m

ut3

pY

JM

Gal3p

Zwf1p

A

Vector control



http://dx.doi.org/10.1101/032136



42

Figure 4 :

III II

I SO

A B

C D



http://dx.doi.org/10.1101/032136



43

Figure 5 :

A B

C D



http://dx.doi.org/10.1101/032136



44

Figure 6 :

A B C

DE



http://dx.doi.org/10.1101/032136



45

Figure 7 :

A B C

DE F



http://dx.doi.org/10.1101/032136



46

Figure 8 :



http://dx.doi.org/10.1101/032136


multiple conformations of gal3 protein drive the galactose ... · mechanism, rather employs a...

Documents